Multi-classes Imbalanced Dataset Classification Based on Sample Information
Chuang Yu, Fengqi Li, Guangming Li, Nanhai Yang · 2015
The classification boundary for multi-classes imbalanced dataset is difficult to judge, posing an important challenge on classification methods. Aiming at this problem, we propose an multi-classes imbalanced data classification algorithm based on sample information. The proposed algorithm applies the sample information measurement to multi-classes imbalanced dataset. Furthermore, a classifier is devised to classify the data. Experiments on IRIS, WINE, GLASS datasets show that our proposed scheme produces a promising result for classifying multi-classes imbalanced data.